bed2gtf

A fast bed-to-gtf converter written in Rust.

translates

chr27 17266469 17281218 ENST00000541931.8 1000 + 17266469 17281218 0,0,200 2 103,74, 0,14675,

into

chr27 stdin gene 17266470 17281218 . + . gene_id "ENSG00000151743"; gene_biotype "protein_coding";

chr27 stdin transcript 17266470 17281218 . + . gene_id "ENSG00000151743"; transcript_id "ENST00000541931.8";

chr27 stdin exon 17266470 17266572 . + . gene_id "ENSG00000151743"; transcript_id "ENST00000541931.8"; exon_number "1"; exon_id "ENST00000541931.8.1";

...

in a few seconds.

Usage

Usage: bed2gtf[EXE] --bed <BED> --isoforms <ISOFORMS> --verbose [verbose]

Arguments:
    --bed <BED>: a .bed file
    --isoforms <ISOFORMS>: a tab-delimited file
    --verbose [verbose]: print verbose [default: true]

Options:
    --help: print help
    --version: print version

Documentation:

Crate:

a .bed file

tab-delimited files with 3 required and 9 optional fields:

chrom   chromStart  chromEnd      name    ...
  |         |           |           |
chr20   50222035    50222038    ENST00000595977.1735    ...

see BED format for more information

a tab-delimited .txt/.tsv/.csv/... file with genes/isoforms:
```
> cat isoforms.txt

ENSG00000198888 ENST00000361390
ENSG00000198763 ENST00000361453
ENSG00000198804 ENST00000361624
```
you can build a custom file for your preferred species using Ensembl BioMart.

Installation

to install bed2gtf on your system follow this steps:

get rust: curl https://sh.rustup.rs -sSf | sh on unix, or go here for other options
run cargo install bed2gtf (make sure ~/.cargo/bin is in your $PATH before running it)
use bed2gtf with the required arguments
enjoy!

Library

to include bed2gtf as a library and use it within your project follow these steps:

include bed2gtf = 1.0.0 under [dependencies] in the Cargo.toml file
the library name is bed2gtf, to use it just write:
```
use bed2gtf::bed2gtf; 
```
or
```
use bed2gtf::*;
```

invoke

let gtf = bed2gtf(bed: PathBuf, isoforms: PathBuf)

Build

to build bed2gtf from this repo, do:

get rust (as described above)
run git clone https://github.com/alejandrogzi/bed2gtf.git && cd bed2gtf
run cargo run --release <BED> <ISOFORMS>(arguments are positional, so you do not need to specify --bed/--isoforms)

Output

bed2gtf will send the output directly to the same .bed file path

.
├── ...
├── isoforms.txt
├── annotation.bed
└── output.gtf

where output.gtf is the result. Any intermediate files are deleted, since there is no need to keep them.

FAQ

Why?

UCSC offers a fast way to convert BED into GTF files through KentUtils or specific binaries (1) + several other bioinformaticians have shared scripts trying to replicate a similar solution (2,3,4).

A GTF file is a 9-column tab-delimited file that holds gene annotation data for a specific assembly (5). The 9th column defines the attributes of each entry. This field is important, as some post-processing tools that handle GTF files need them to extract gene information (e.g. STAR, arriba, etc). An incomplete GTF attribute field would probably lead to annotation-related errors in these software.

Of the available tools/scripts mentioned above, none produce a fully functional attribute GTF file conversion. (1) uses a two-step approach (bedToGenePred | genePredToGtf) written in C, which is extremely fast. Since a .bed file does not preserve any gene-related information, this approach fails to a) include correct gene_id attributes (duplicated transcript_ids) b) append 3rd column gene features.

This is an example:

chr27 stdin transcript 17266470 17281218 . + . gene_id "ENST00000541931.8"; transcript_id "ENST00000541931.8";

chr27 stdin exon 17266470 17266572 . + . gene_id "ENST00000541931.8"; transcript_id "ENST00000541931.8"; exon_number "1"; exon_id "ENST00000541931.8.1";

On the other hand, available scripts (2,3,4) fall into bad-formatted outputs unable to be used as input to other tools. Some of them show a very customed format, far from a complete GTF file (2):

chr20 ---- peak 50222035 50222038 . + . peak_id "chr20_50222035_50222038";

chr20 ---- peak 50188548 50189130 . + . peak_id "chr20_50188548_50189130";

and others (4) just provide exon-related information:

chr20 ensembl exon 50222035 50222038 . + . gene_id "ENST00000595977.1735"; transcript_id "ENST00000595977.1735"; exon_number "0

chr20 ensembl exon 50188548 50188930 . + . gene_id "ENST00000595977.3403"; transcript_id "ENST00000595977.3403"; exon_number "0

This is where bed2gtf comes in: a fast and memory efficient BED-to-GTF converter written in Rust. In ~15 seconds this tool produces a fully functional GTF converted file with all the needed features needed for post-processing tools.

How?

bed2gtf takes advantage of UCSC binaries. These executables show slightly lower computation times than those seen in a similar rust implementation. This tool uses bedToGenePred + genePredToGtf as a first step, automates their download and updates permissions. After that, this tool overcomes the limitations of gene_id and gene-feature fields by expecting a tab-delimited isoforms file. Each unique transcript seen in the new gtf file is mapped to their respective gene_id. With this information, bed2gtf simply constructs gene-feature lines and corrects gene_ids.

Limitations

At the time of bed2gtf being publicly available some gaps have not been covered yet.

Sorting. This tool does not implement a sorting step. While it is not mandatory, and probably would not affect post-processing tools, it could increase computation time. A sorting step should consider a) chromosome, b) boundaries (start-end) and c) features (gene > transcript > exon, start_codon, CDS, stop_codon, UTR).
Time. Although bed2gtf is pretty fast compared to other implementations (e.g python3; unpublished data), I personally think that this tool could achieve even faster computation times with more detailed algorithms and structures.
Biotype. As you may know (or not), GTF files specify the gene_biotype of each entry (e.g. protein_coding, processed_pseudogene, snoRNA, etc). This is probably the biggest limitation in this release. Currently, bed2gtf assumes all genes/transcripts provided are protein_coding. In future releases will probably be an option to specify the gene_biotype [-b/--biotype]. This feature could also be extracted from Ensembl BioMart and append it to the isoforms file; however, map them to each line will probably cost efficiency without a better framework.

bed2gtf 1.0.0